长江中下游4个湖泊非色素颗粒物吸收系数光谱模型

利用指数模型和幂函数模型对2007年和2010年长江中下游4个湖泊58个样本 a_d(λ)(非色素颗粒物吸收系数) 光谱进行拟合,以 R2(决定系数)、RE(相对误差)和RMSE(均方根误差)等统计参数判定模型效果. 指数模型和幂函数模型 R2平均值、RE、RMSE分别为0.998、7.01%、0.015 m-1和0.994、15.90%、0.027 m-1. 统计检验显示,指数模型的 R2显著大于幂函数模型(ANOVA,P<0.001),而RE和RMSE则显著小于幂函数模型(ANOVA,P<0.001). 指数模型能更准确地拟合 a_d(λ),得到的 S_d(光谱斜率)平均值为(12.21±1.08)μm-1,400~700 nm波段内变异系数为8.85%,空间上4个湖泊之间变化不是很明显,仅傀儡湖的 S_d略低于其他3个湖泊. a_d(λ)与ρ(ISM)(无机悬浮颗粒物浓度)呈极显著正相关,通过ρ(ISM)可以得到长江中下游典型浅水湖泊 a_d(λ)光谱模型. 

Spectral Absorption Models of Tripton in Four Lakes in the Middle and Lower Reaches of the Yangtze River

The spectral absorptions of non-pigment particulate matter in Lake Kuileihu, Lake Donghu, Lake Liangzi and Lake Honghu were measured on 58 samples collected from September-October 2007 and October 2010. The goal was to develop exponential and power models which included the parameters of spectral slope ( S_d), inorganic suspended matter and wavelengths describing the spectral absorption of non-pigment particulate matter. The determination coefficient (R2), relative error (RE) and root mean square error (RMSE) were used to assess the performance of the models. The average R2 value of the exponential model was 0.998, which was higher than that of the power model (0.994) (ANOVA, P<0.001). However, RE and RMSE for the exponential model (7.01% and 0.015 m-1, respectively) were lower than those for the power model (15.90% and 0.027 m-1, respectively) (ANOVA, P<0.001). These results indicated that the exponential model was better than the power model. The mean S_d value of non-pigment particulate matter from the exponential model was (12.21±1.08)μm-1, with a coefficient of variation of 8.85% for the range of 400-700 nm. The S_d in Lake Kuileihu was slightly higher than those of the other three lakes, but without a significant difference. This suggests that a mean value could be used to model non-pigment particulate matter spectral absorption in the four lakes. Significant linear correlations were found between non-pigment particulate matter absorption coefficient and inorganic suspended matter. The results indicate the potential to develop regional models to predict the absorption spectra using inorganic suspended matter concentration.